I'd like to point out that the message below was sent
to Westnet sites (Gordon Cook forwarded it to NANOG without taking
the context into consideration). From our perspective SprintLink
and MCI were in fact partitioned. My customers probably don't care
that other SprintLink sites could get to MCI customers (except in
the "gee this Internet thing is good for the world" sort of way).
Last night about 7pm Mountain time we lost connectivitiy from
one of our monitoring stations to all our "SprintLink only" sites. This
monitoring station uses MCI as it's NSP. After doing some traces it
was pretty clear that MCI was blackholing traffic that came from the
Denver MCI POP headed to SprintLink sites. It was bouncing around
between MCI's SF nodes. The MCI NOC was in fact aware of the problem
and told me it was at escalation level 4. I was told this morning
by the MCI NOC that MCI had an "SSE hang on a border router" and that
a reset had fixed the problem. MCI ticket #2778.
We regained connectivity at about 7:30pm, so it lasted just
under half an hour for us.
I doubt the nanog list is the place for this though.